| 英文摘要 | Unsupervised hyperspectral unmixing (i.e., UHU) is one of the hottest topics in the remote sensing image processing, which is one of the most important tools to quantitatively analyze the hyperspectral images from remote sensing. It plays a fundamental role in a wide range of applications, such as hyperspectral visualization and understanding, hyperspectral compression and reconstruction, detection and identification substances in the scene, hyperspectral enhancement and high-resolution hyperspectral imaging etc. This task is still highly challenging due to the following three issues: (1) since both endmember and abundance are unknown, the solution space for the UHU models is really large; (2) it is very easy for hyperspectral images to be badly degraded by various kinds of noises, resulting in many outlier channels; (3) when there is no prior knowledge on hyperspectral images, the notation of pure material (i.e., endmembers) is subjective and problem dependent. To address the above the above three issues, reasonable prior knowledge is proposed to restrict the solution space, or even to bias the solution toward good stationary points. Then, various robust measures are employed for the representation loss, preventing large errors from dominating our objective. Besides, we propose an accelerated robust subset selection method and elaborate its applications for the UHU task. The main contributions of this paper are summarized as follows: 1. We propose a structured sparse regularized method, for the UHU problem, from two aspects. First, we incorporate a graph Laplacian to encode the manifold structures embedded in the hyperspectral data space. In this way, the highly similar neighboring pixels can be grouped together. Second, the lasso penalty is employed for the fact that pixels in the same manifold structure are sparsely mixed by a common set of relevant bases. These two factors act as a new structured sparse constraint. With this constraint, our method can learn a compact space, where highly similar pixels are grouped to share correlated sparse representations. 2. We propose a UHU method via Data-guided Sparsity. To reduce the solution space, many methods have been proposed by exploiting various priors. In practice, these priors would easily lead to some unsuitable local minima. This is because they are achieved by applying an identical strength of constraint to all the factors, which does not hold in practice. To overcome this limitation, we propose a ... |
修改评论